Metric Nearness Made Practical
Authors: Wenye Li, Fangchen Yu, Zichen Ma
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In empirical evaluations, the proposed approach runs at least an order of magnitude faster than the state-of-the-art solutions, with significantly improved scalability, complete conformity to constraints, less memory consumption, and other desirable features in real applications. |
| Researcher Affiliation | Academia | Wenye Li1,2, Fangchen Yu1, Zichen Ma1 1 The Chinese University of Hong Kong, Shenzhen 2 Shenzhen Research Institute of Big Data 2001 Longxiang Boulevard, Longgang District, Shenzhen, China wyli@cuhk.edu.cn, fangchenyu@link.cuhk.edu.cn, zichenma1@link.cuhk.edu.cn |
| Pseudocode | Yes | Algorithm 1: The Proposed HLWB Algorithm |
| Open Source Code | No | The paper refers to existing implementations of other algorithms (e.g., "2https://optml.mit.edu/work/soft/metricn.html", "3https://github.com/rsonthal/Project And Forget", "4https://github.com/spitis/deepnorms") but does not provide a link or explicit statement about the availability of the code for *their own* proposed methodology. |
| Open Datasets | Yes | Besides, the real MNIST dataset (Le Cun et al. 1998)7, which consists of grayscale images of hand-written digits, was used in the experiment. |
| Dataset Splits | No | The paper mentions generating artificial datasets and using the MNIST dataset, but it does not specify the train/validation/test splits, only that "n = 100/500/1,000/1,500 nodes were artificially generated" and "n varying from 100 to 1,500" for MNIST. |
| Hardware Specification | Yes | Most experiments were executed on a conventional server with a single CPU (intel Xeon 8180) enabled, except the Deep Norm algorithm that ran on a deep learning platform. |
| Software Dependencies | No | The paper mentions specific solvers like CPLEX and MOSEK, but does not provide their version numbers. It also links to open-source implementations of compared algorithms, but without stating specific software versions for those. |
| Experiment Setup | Yes | The measurement matrix Do was formed by adding random noise to the elements of Dg, with each do ij = max{0, dg ij + ζ · mean(Dg) · N(0, 1)} where ζ = 0.5 and 0.8 respectively and mean(Dg) denotes the mean value of all entries of Dg. |